Authors: Reda ALAMI
The ML-kNN algorithm is one of the most famous and most efficient multi-label classifier. Its performances are very remarkable when compared with the other state-of-art multi-label classifiers. Nevertheless, it suffers from two major drawbacks: its accuracy crucially depends on the metric function used to compute distances between instances, and when dealing with high dimensions data, the neighborhoods identification task becomes very slow. So, both metric learning and dimensionality reduction are essential to improve the ML-kNN performances. In this report, we propose a novel multi-label Mahalanobis distance learned via a supervised dimensionality reduction approach that we call ML-ARP. ML-ARP is a process that adapts random projections on a multi-label dataset to improve the ML-kNN performances. Unlike most state of art multi-label dimensionality reduction approaches that solve eigenvalue or inverse problem, our method is iterative and scales up with high dimensions. There is no eigenvalue or inverse problems to solve. Experiments show that the ML-ARP allows us to highly upgrade the ML-kNN classifier. Statistical tests assert that the MLARP is better than the remaining state-of-art multi-label dimensionality reduction approaches
Comments: 75 Pages.
[v1] 2019-06-21 13:51:15
Unique-IP document downloads: 2 times
Vixra.org is a pre-print repository rather than a journal. Articles hosted may not yet have been verified by peer-review and should be treated as preliminary. In particular, anything that appears to include financial or legal advice or proposed medical treatments should be treated with due caution. Vixra.org will not be responsible for any consequences of actions that result from any form of use of any documents on this website.
Add your own feedback and questions here:
You are equally welcome to be positive or negative about any paper but please be polite. If you are being critical you must mention at least one specific error, otherwise your comment will be deleted as unhelpful.